I would second this. Get a 10053 trace from both versions and look at the trace file in your favorite graphical diff tool. Look for the cardinality/density calculation. I would do this for the full query as well as your one table test case . What are the two execution plans of the full query? On Mon, Nov 23, 2009 at 3:55 PM, Stefan Knecht <knecht.stefan@xxxxxxxxx> wrote: > Just a quick thought... > > One thing you can try is to take a 10053 trace, and check for the values of > the various underscore (optimizer features), as well as bugfixes listed in > the tracefile, for both values for optimizer_features_enable. Any difference > you may spot may lead you to some more information. -- Regards, Greg Rahn http://structureddata.org -- //www.freelists.org/webpage/oracle-l